Automatic versus morphophonemic rules: some Old French evidence

Abstract

Rule taxonomy has been the focus of much attention in the recent phonological literature. In particular, one approach (Zwicky 1984) has argued that the appropriate division of rules is into automatic versus nonautomatic types rather than the 'traditional' morphophonemic versus allophonic classification. Based on evidence from the history of French which shows a division within a class of forms subject to an automatic rule between those which are in a morphophonemic relationship and those which are not, this short paper will conclude that a three-way classification, nonautomatic morphophonemic, automatic morphophonemic, and allophonic, is necessary for a proper understanding of (morpho)phonological behavior.In the recent phonological literature, a significant amount of attention has focused on questions of rule taxonomy: what are the different types of phonological rules, what properties unite or divide them, what are their possible interactions? As a by no means exhaustive sample of types, we may list allophonic versus morphophonemic rules, processes versus rules, phonological rules proper, sandhi rules, via rules, lexical versus postlexical rules. Relevant distinguishing properties have included the presence or absence of nonphonological information in the structural description, phonotactic conditioning, possibility of exceptions, variability, neutralizing versus nonneutralizing, structure-changing versus structure-preserving, automatic versus nonautomatic, inversion, feature changing versus feature inserting, or phonetic naturalness, among others. Possible interactions involve the grouping of rules into ordered blocks (typically nonphonological preceding phonological, under one of the definitions of this distinction), cyclicity, the possibility of rule insertion, or the ubiquitous question(s) of ordering. 1 A glance at recent publications would permit immediate expansion of this inventory. Linguistics 25 (1987), 915-923